Global Pharmacy Counseling and Support System, Pharmacy Counseling and Support Methods, and Programs

Abstract

This system leverages the expertise of pharmacists worldwide without geographical constraints to provide safe, multilingual support for consultations regarding prescription drugs, health, and cosmetic ingredients, and includes an adaptive reinference mechanism and an audit log mechanism based on the reliability of AI processing results. [Solution] In a system comprising a client terminal, a pharmacist terminal, and a server device, the server device receives various information from the client terminal via symptom information receiving means, drug image data receiving means, and drug information receiving means. It also includes an AI screening means equipped with an adaptive reinference mechanism based on reliability, a drug identification means using image recognition and reliability calculation, an interaction check means equipped with difference evaluation and uncertainty indicators, a 24-hour AI chat assistant means, a multilingual translation means that supports specialized terminology, a consultation summarization means, and an audit log means that ensures traceability of all processing. All AI processing results are provided to the client after final confirmation by a pharmacist, ensuring safety.

Description

Technical Field

[0001] The present invention relates to a global pharmacy consultation support system, a pharmacy consultation support method, and a program that utilize artificial intelligence (AI) technology to connect pharmacists and consultants worldwide online and support consultations on prescription drugs, health, and cosmetic ingredients in multiple languages.

Background Art

[0002] In recent years, in developed countries around the world, so-called "medical collapse" is progressing due to soaring medical costs, doctor shortages, and the breakdown of insurance systems. Patients cannot easily visit medical institutions and are losing places to casually consult about minor concerns regarding medicine and health.

[0003] On the other hand, although pharmacists have advanced specialized knowledge, in many countries, their utilization is limited to dispensing operations, and the current situation is that their knowledge and experience are not fully returned to society.

[0004] Also, when traveling overseas or working abroad, the need to obtain drugs equivalent to those prescribed in one's home country locally is increasing. However, drug product names vary from country to country, and there is a problem that it is difficult to identify the name even if the active ingredients and contents are the same.

[0005] Furthermore, regarding the risk of interactions (drug combinations) when taking multiple drugs simultaneously, the means by which ordinary consumers can easily confirm are limited. In particular, checking for interactions between drugs prescribed by different medical institutions is an important issue.

[0006] In addition, with globalization, language barriers have become a major obstacle in medical consultations, and the difficulty of consulting about medicine and health in languages other than one's mother tongue has been pointed out.

[0007] Conventional online medical consultation systems lacked a feedback mechanism to quantitatively calculate the reliability of AI processing results and to request additional information from the user to re-infer results if the reliability was insufficient. Furthermore, they lacked a traceability mechanism to consistently record and store the input, output, and user responses of the AI ​​processing as audit logs. [Prior art documents] [Patent Documents]

[0008] [Patent Document 1] Patent No. 7481053 [Patent Document 2] Patent No. 7274529 [Non-patent literature]

[0009] [Non-Patent Document 1] Ministry of Health, Labour and Welfare, "Guidelines on the Safe Use of Pharmaceuticals and Medical Devices" [Overview of the Initiative] [Problems that the invention aims to solve]

[0010] The drug support system described in Patent Document 1 is a system that provides drug-related support between pharmacists and patients, but it does not include a multinational equivalent drug identification function, an AI image recognition drug identification function, an adaptive reinference mechanism based on reliability, or an audit log mechanism. The drug identification system described in Patent Document 2 is a system that identifies drugs based on drug images, but it does not include a function to generate multinational equivalent drug information based on the identification results, a drug interaction check function, a multilingual translation function, or a matching function with pharmacists.

[0011] Given the above background, the following challenges exist: (1) There is no system to utilize pharmacists' expertise globally without geographical constraints. (2) There is no means to quickly identify equivalent drugs of domestic drugs overseas from a single photograph across multiple countries. (3) There is no integrated system that allows consumers to easily check the risk of drug interactions between multiple drugs and consult with a pharmacist as needed. (4) There is no technology to support drug consultations in multiple languages ​​in real time while maintaining the accuracy of technical terms. (5) There is no system to automate the summarization of consultation content and extraction of important points to improve the quality and efficiency of consultations. (6) There is no adaptive system to quantitatively evaluate the reliability of AI processing results and collect additional information and re-infer if it is insufficient. (7) There is no system to record and save the entire process of AI processing as a traceable audit log. (8) There is no system to automatically detect highly urgent symptoms when symptoms are entered and quickly guide users to make emergency calls. (9) There is no system to check for dangerous interactions not only between drugs but also between drugs and food or beverages in advance. (10) There is no system that allows users who have difficulty typing text to input symptom information by voice.

[0012] To solve the above problems, the present invention aims to provide a platform that securely connects pharmacists and clients globally by integrating multiple AI functions utilizing artificial intelligence technology, and incorporating an adaptive reinference mechanism based on reliability and an audit log mechanism. [Means for solving the problem]

[0013] To solve the above problems, the present invention provides a global drug consultation support system comprising a consultation terminal, a pharmacist terminal, and a server device connected to the consultation terminal and the pharmacist terminal via a network, wherein the server device includes: symptom information receiving means for receiving symptom information transmitted from the consultation terminal by the consultation user; AI screening means for analyzing the symptom information using an AI model employing natural language processing to determine the optimal consultation category and recommended pharmacist; drug image data receiving means for receiving drug image data transmitted from the consultation terminal; drug identification means for extracting drug information from the drug image data using an image recognition AI model and generating information on equivalent drugs in multiple countries based on the extracted drug information; drug information receiving means for receiving information on drugs currently being taken transmitted from the consultation terminal; interaction checking means for analyzing drug interactions using an AI model employing natural language processing to determine risk levels; and multilingual translation means for translating messages transmitted and received in consultation communication between the consultation terminal and the pharmacist terminal between multiple languages.

[0014] Furthermore, the present invention provides a drug consultation support method that is executed by a server device connected via a network to the consultation terminal and the pharmacist terminal.

[0015] Furthermore, the present invention provides a program for causing a computer to function as one of the means in the global drug consultation support system.

[0016] In particular, the three key technical features of this invention are as follows: Firstly, it is based on a safety design philosophy that positions the AI ​​not as a decision-maker, but as a "support wheel" that safely guides the user to consultation, and requires final confirmation by a pharmacist for all AI processing results. Secondly, it is equipped with an adaptive feedback mechanism that quantitatively calculates the reliability of the processing result in each AI processing module, and sends additional confirmation questions to the user's consultation terminal and performs re-inference if the reliability is below a predetermined threshold. Thirdly, it ensures traceability by recording and saving all AI processing input data, processing results, reliability, user responses, and pharmacist's final confirmation as audit logs. [Effects of the Invention]

[0017] The present invention provides the following benefits: (1) High-quality drug consultations are possible without geographical constraints by utilizing the expertise of pharmacists worldwide online. (2) Equivalent drugs in multiple countries can be instantly identified from a single photo of a drug, supporting drug acquisition when traveling abroad. (3) Drug interaction risks are visualized in five stages, supporting safe medication use by consumers. (4) A translation function supporting five languages ​​enables drug consultations that transcend language barriers. (5) The AI ​​summarization function streamlines the recording and sharing of consultation content. (6) Medical safety is ensured by a safety design in which AI focuses on judgment assistance and pharmacists make the final decisions. (7) Adaptive reinference based on reliability improves the accuracy of AI processing and user confidence. (8) Audit logs ensure traceability of all processing, supporting quality control and legal compliance. (9) Emergency sign detection automatically detects highly urgent symptoms and guides users to make quick emergency calls, ensuring user safety. (10) By checking for drug-food or beverage interactions, life-threatening combinations can be warned in advance. (11) The voice input function provides a convenient consultation method for users who have difficulty typing text. [Brief explanation of the drawing]

[0018] [Figure 1] Figure 1 is a block diagram showing the overall configuration of a global drug consultation and support system according to one embodiment of the present invention. [Figure 2] Figure 2 is a flowchart showing a consultation reservation flow according to an embodiment of the present invention. [Figure 3] Figure 3 is a flowchart showing the flow of medicine identification AI processing (including confidence branching) according to an embodiment of the present invention. [Figure 4] Figure 4 is a flowchart showing the flow of interaction check processing according to an embodiment of the present invention. [Figure 5] Figure 5 is a diagram showing AI function cooperation and data flow according to an embodiment of the present invention. [Figure 6] Figure 6 is a flowchart showing the flow of multilingual translation processing according to an embodiment of the present invention. [Figure 7] Figure 7 is a diagram showing screen transition according to an embodiment of the present invention.

Embodiments for Carrying out the Invention

[0019] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. Note that the following embodiments do not limit the present invention, and various modifications are possible within the scope of the gist of the present invention. <Definition of Terms>

[0020] In this specification, the following terms are used in the following meanings. "User terminal" refers to a terminal (such as a smartphone or a PC) operated by a counselor or a user. "Server device" refers to an API / application server, which is a device that executes inference, verification, presentation control, etc. "Reliability" refers to a numerical value (e.g., in the range of 0.0 to 1.0) representing the validity of an estimation result by an AI model, and is used synonymously with "certainty". "Candidate information" refers to candidates (which may include multiple ones) as estimation results. <Overall System Configuration>

[0021] FIG. 1 is a block diagram showing the overall configuration of a global pharmaceutical consultation support system (hereinafter also referred to as "this system") according to an embodiment of the present invention. This system includes a counselor terminal (10), a pharmacist terminal (20), and a platform server (100), which are connected to each other via the Internet so as to be communicable.

[0022] The counselor terminal (10) is an information terminal such as a smartphone (11) or a PC terminal (12), and has a camera function (13). The counselor can input symptoms, take and upload pictures of drugs, make a consultation reservation with a pharmacist, conduct a video call, etc. via the counselor terminal (10).

[0023] The pharmacist terminal (20) includes a management screen (21), a consultation response screen (22), and a video call (23). The pharmacist can approve or reject a consultation reservation, exchange messages with the counselor, respond to consultations via video calls, and finally confirm the AI processing results, etc. via the pharmacist terminal (20).

[0024] The platform server (100) includes a front-end part (110) and a back-end part (120). The front-end part (110) is implemented as a web application and provides a user interface for counselors and pharmacists. The back-end part (120) is implemented as an API server and controls the AI processing module group (200), accesses the database (400), and coordinates with external services, etc. <Configuration of the AI Processing Module Group>

[0025] The AI processing module group (200) is composed of the following six modules. Each module entrusts the processing to a natural language processing AI model or an image recognition AI model via the external AI API cooperation part (300) from the back-end part (120). [[ID=,19]] In this specification, the AI ​​screening module (210), the 24-hour AI chat assistant module (220), the drug identification AI module (230), the interaction check module (240), the AI ​​translation module (250), and the AI ​​summary module (260) may be abbreviated in the drawings as AI screening (210), AI chat assistant or 24-hour AI chat (220), drug identification AI (230), interaction checker (240), AI translation (250), and AI summary (260), respectively. (1) AI screening module (210)

[0026] The AI ​​screening module (210) receives symptom text entered by the user in the screening section of the home screen via a symptom information reception mechanism. The received symptom text is sent to a natural language processing AI model for analysis. As a result of the analysis, optimal consultation type candidates (three categories: prescription drug consultation, health consultation, and cosmetic ingredient consultation) and the confidence level of each candidate are generated.

[0027] If the confidence level is below a predetermined first threshold, this module sends additional confirmation questions (such as whether the user is taking any medications, details of the purpose of the consultation, urgency, age group, and medical history) to the user's terminal (10) and performs reclassification based on the answers. If the confidence level is below a predetermined second threshold, the module suppresses the transmission of recommendation results and directs the user to a specialist or issues a warning.

[0028] Based on the confirmed consultation type, candidate pharmacists are extracted from the pharmacist database. For the extracted candidates, a comprehensive score is calculated by weighting multiple objective indicators, including evaluation indicators (review evaluation, etc.), consultation performance indicators (number of cases handled, completion rate, etc.), and consultation type matching indicators (degree of matching of specialty), using weights dynamically adjusted according to reliability, urgency, and the user's desired conditions. The top N pharmacists are ranked and sent to the consultation terminal. When sending, the contribution ratio of each indicator to each candidate's comprehensive score is also displayed as supporting information. (2) 24-hour AI chat assistant module (220)

[0029] The 24-hour AI chat assistant module (220) corresponds to the AI ​​chat assistant means in the claims and operates via a chat widget that floats in the lower right corner of the screen on all pages. It uses a natural language processing AI model to respond to questions about medicine and health 24 hours a day. The module suggests consultation types based on symptoms, recommends pharmacists, and directs users to profile pages. Furthermore, it includes a quick reply button function, a conversation history retention function, 5-language support, and a fallback function. The module is provided as the AI ​​chat (P800) shown in Figure 7 or as a 24-hour AI chat widget that floats across the entire screen. (3) Drug Identification AI Module (230): Universal Drug Identifier

[0030] The drug identification AI module (230) is one of the core functions of the present invention. When a user uploads a photo of a drug by dragging and dropping or taking a picture with a camera, the image data is received via the drug image data receiving means, converted to a predetermined format, and then sent to the image recognition AI model. In the drug identification AI processing flow shown in Figure 3, the user takes or uploads an image of the drug in step S200, image preprocessing (resizing and conversion to a predetermined format) is performed in step S210, and the image is sent to the image recognition AI model in step S220. The drug identification AI module (230) is available for use in the drug identification AI (P500) shown in Figure 7.

[0031] The image recognition AI model extracts image features, including color, shape, markings, and packaging information, from an image to generate candidate drug information (drug name, active ingredient, content, manufacturer, etc.), and calculates a confidence score based on at least one of the degree of matching of image features, the degree of matching of marking recognition results, and the degree of matching of packaging type. In step S230, the image recognition AI model performs AI image recognition and drug information extraction, and candidate drug information including drug name, active ingredient, content, and manufacturer is generated.

[0032] If the confidence level is below a predetermined threshold, this module sends additional confirmation questions (such as inputting engraved text, requesting re-shooting, adding images from multiple angles, or inputting text from the packaging label) to the user's terminal and performs re-inference based on the answers sent from the user's terminal. If multiple drug candidates exist, they are sent to the user's terminal in order of priority, and it is also possible to regenerate equivalent drug candidate information based on the user's selection.

[0033] Based on the confirmed drug information, information on equivalent drugs in multiple countries (Japan, the United States, the United Kingdom, Germany, France, South Korea, China, Australia, Canada, Thailand, Singapore, India, Spain, etc.) is generated. In selecting candidate equivalent drugs, the selection criteria are displayed based on at least one of the following: matching active ingredients, matching or proximity of doses, matching dosage form, and matching therapeutic classification. Furthermore, as shown in Figure 3, the target country is selected in step S240, equivalent drug data for each country is generated in step S250, and the results, including drug name, availability, price range, and precautions, are displayed in step S260.

[0034] Dose conversions are performed based on different unit systems or standards in each country, and if there are differences in dosage or dosage form, warning information and the basis for the conversion are added. Equivalent drug information includes the drug name in each country, the availability of generic drugs, availability (high / medium / low), estimated price range, regulatory classification information (OTC / prescription, etc.), and precautions. Regulatory classification information is added based on country-specific regulatory tables or country-specific regulatory APIs. (4) Interaction check module (240)

[0035] The interaction check module (240) functions on the medication record page. Based on the medication information received via the medication information reception means, it retrieves existing drug groups with a "currently taking" flag set to true from the database and sends them to a natural language processing AI model to analyze drug interactions. In the interaction check processing flow shown in Figure 4, the list of currently taking medications is obtained in step S300, new drug additions are accepted in step S310 if necessary, and the drug list is sent to the natural language processing AI model in step S320. The interaction check module (240) is available in the medication record (P600) and interaction (P700) shown in Figure 7.

[0036] In particular, when a new drug is added, the interaction between the new drug and the existing group of drugs being taken is evaluated by differential assessment. That is, only interaction pairs that newly arise due to the newly added drug are extracted and displayed. In addition to the "currently taking" flag, it is also possible to obtain a list of existing drugs by considering the overlap of taking periods based on the start and end dates.

[0037] The analysis results are determined as one of five risk levels: Critical, High, Moderate, Low, or None. The risk level classification is based on at least one of the following: contraindications of the interaction, the need for dose adjustment, and the degree of caution required for concomitant use. In addition, an uncertainty index is generated, and if the uncertainty index is above a predetermined threshold, cautionary information is highlighted. Step S330 performs AI interaction analysis, and step S340 performs risk level determination.

[0038] Interaction assessment can utilize estimations based on language models in addition to matching results from external interaction databases or rule engines. Recommended responses include at least one of the following: recommendation to see a doctor, consultation on discontinuing medication, consultation on dosage adjustment, adjustment of administration timing, and presentation of observation items. If the risk level exceeds a predetermined threshold, a notification is generated prompting the user to consult a pharmacist, and further notifications can be sent via in-app or email. As shown in Figure 4, step S350 displays the results and prompts the user to consult a pharmacist. (5) AI translation module (250)

[0039] The AI ​​translation module (250) functions within the message tab on the consultation details page. Using a natural language processing AI model, it translates messages within the consultation between five languages: Japanese, English, Korean, Chinese, and Spanish. To ensure accuracy in translation, the system employs specialized prompt control for pharmaceutical terminology, cosmetic ingredient names, and health-related terms. In the multilingual translation processing flow shown in Figure 6, the consultation message is sent in step S400, the translate button is pressed and the target language is selected in step S410, a translation request is sent to the natural language processing AI model in step S420, the AI ​​translation process, including pharmaceutical terminology handling, is executed in step S430, and the translation result is displayed in step S440. (6) AI Summary Module (260)

[0040] The AI ​​summary module (260) corresponds to the AI ​​summary means in the claim and functions in the "AI Summary" tab of the consultation details page after the consultation status becomes "Completed". Using a natural language processing AI model, it analyzes all messages exchanged during the consultation and automatically generates a summary of the consultation content, key points, and advice / recommendations. <Audit logging mechanism>

[0041] This system is equipped with an audit log means (270) that records and stores, along with the processing time, various information transmitted from the client terminal, AI processing results, reliability (if calculated), presented content transmitted to the client terminal or pharmacist terminal, user selections or additional responses, and the pharmacist's final confirmation for all six AI modules mentioned above. This ensures traceability of all AI processing and is used for quality control, legal compliance, and post-processing verification. <Consultation Flow>

[0042] Figure 2 shows the consultation reservation flow in this system. This system is equipped with a consultation flow management function corresponding to the consultation management means in the claim, and the consultation flow consists of the following steps S100 to S180.

[0043] Step S100 (Category Selection): The person seeking advice selects their topic from four categories: prescription drugs / drug interactions, side effects, health consultations, and cosmetic ingredients.

[0044] Step S110 (AI Symptom Screening): The AI ​​screening module (210) analyzes the symptom text received via the symptom information reception means and generates candidate consultation types and confidence levels. If the confidence level is below a threshold, additional confirmation questions are sent to the consultation terminal, and reclassification is performed.

[0045] Step S120 (Consultation Type / Pharmacist Recommendation): Based on the finalized analysis results, a list of optimal pharmacists ranked by recommended consultation category and overall score, along with the rationale for the ranking, is sent to the client's terminal. At this time, the pharmacist list (P100) shown in Figure 7 is displayed on the client's terminal.

[0046] Step S130 (Inputting Consultation Details): The person seeking consultation enters detailed consultation details, country / region, language used, and urgency level (normal / urgent). The person seeking consultation refers to the pharmacist details shown in Figure 7 (P200) as needed, and then enters detailed consultation details in the consultation reservation (P300).

[0047] Step S140 (Date and Time Selection & Booking Confirmation): The client selects the date, time, and time slot for their consultation using the calendar UI and confirms the booking.

[0048] Step S150 (Pharmacist Approval): The selected pharmacist reviews the reservation details and makes a decision to approve or reject it.

[0049] Step S160 (Payment Processing): After approval by the pharmacist, online payment is processed via the payment API.

[0050] Step S170 (Video Consultation): The consultant and pharmacist conduct a real-time consultation using the video call function. During this time, message translation by the AI ​​translation module (250) is available. While the consultation is in progress, the consultant and pharmacist can use message exchange and video calls as shown in the consultation details (P400) in Figure 7.

[0051] Step S180 (AI Summary Generation): After the consultation is completed, the AI ​​summary module (260) automatically summarizes the consultation content and generates key points and advice. After the consultation is completed, the AI ​​summary is displayed on the consultation details (P400) or a related screen.

[0052] The inputs, outputs, and user selections for the AI ​​processing in all the steps described above are recorded by the audit logging device (270). <Other components>

[0053] The database (400) manages user information, pharmacist profiles, consultation history, medication record data, etc.

[0054] The payment processing unit (500) works in conjunction with the payment API to process online payments for consultation fees.

[0055] The video call processing unit (600) enables real-time video calls.

[0056] The drug database (700) stores drug information, generic drug information, regulatory classification information, and drug interaction data from multiple countries. <Emergency Sign Detection Mechanism>

[0057] The AI ​​screening module (210) of this system further includes an emergency sign detection unit. The emergency sign detection unit matches the symptom text received via the symptom information receiving means with a predetermined emergency symptom pattern (more than 12 categories including chest pain, stroke signs, shortness of breath, massive bleeding, seizures, anaphylaxis, overdose, etc.) and determines whether or not there is an emergency sign.

[0058] If an emergency sign is detected, the emergency sign detection unit generates emergency warning information, including the details of the detected emergency symptom, a message indicating the urgency level, and recommended actions, and sends it to the caller terminal (10). The emergency warning information includes the emergency telephone number for the country where the caller terminal is located or the country information set by the user. This function is positioned as providing safety information, not as a triage determination.

[0059] Furthermore, the quick consultation page features an immediate emergency keyword detection function on the client side. This function instantly matches keywords with red flag patterns in five languages ​​without API communication, and automatically turns on the urgency flag and displays a warning upon detection. <Mechanism for checking drug-food or beverage interaction>

[0060] The interaction check module (240) of this system further includes a drug-food interaction analysis unit. In addition to checking interactions between drugs, the drug-food interaction analysis unit analyzes dangerous combinations of drugs with food or beverages.

[0061] The food categories to be analyzed include more than 10 categories, such as grapefruit, vitamin K-containing foods, tyramine-containing foods, dairy products, alcohol, caffeine, potassium-containing foods, licorice, high-fiber foods, and pomelo / Seville oranges. For each food category, the risk of interaction with medications being taken is evaluated, and the severity (critical / high / moderate) and the presence or absence of emergency transport risk are determined.

[0062] The evaluation results are sent to the user's terminal (10) as drug-food interaction information, including the drug name, foods or beverages to avoid, the reason for the interaction, and countermeasures. Combinations that pose a risk of emergency transport are highlighted with a warning. This function also operates similarly during pre-checks when adding new drugs. <Voice Input Mechanism>

[0063] The symptom information reception mechanism of this system has a voice input function in addition to text input. The voice input function converts voice data acquired via the microphone function of the user's terminal (10) into text using speech recognition processing such as the Web Speech API, and accepts the converted text as symptom information.

[0064] The voice input function supports symptom text input in the AI ​​screening module (210), question input in the AI ​​chat assistant module (220), and message input in the translation chat panel during video calls. This improves convenience for users who have difficulty with text input (elderly people, visually impaired people, etc.). Voice input supports five languages ​​and includes a control to prevent accidental transmission when IME conversion is confirmed. <Setting the threshold>

[0065] The thresholds for each AI processing module in this system are set in the following manner. The thresholds for the confidence level (a numerical value in the range of 0.0 to 1.0) output by each AI processing module (AI screening module (210), drug identification AI module (230), etc.) are set in advance by the system designer based on evaluation of recognition accuracy using test data, analysis of the misjudgment rate, and operational accuracy verification considering user safety.

[0066] For example, in the drug identification AI module (230), if the confidence level of the image recognition result is less than a predetermined first threshold (e.g., 0.7), an additional confirmation question is sent to the user's terminal. In the AI ​​screening module (210), if the confidence level of the consultation type candidate is less than a predetermined first threshold, an additional confirmation question is sent and reclassification is performed. Furthermore, if the confidence level is less than a predetermined second threshold (e.g., 0.3), the transmission of recommendation results is suppressed, and direct guidance to an expert or a warning is issued. The specific values ​​of the thresholds can be adjusted as appropriate even after the system is put into operation, based on user feedback and statistical analysis of processing results. <Safety design philosophy>

[0067] In this invention, AI is not positioned as a decision-maker, but rather as a "support wheel" that safely guides the user towards consultation. In other words, all six AI functions are used for organizing information, translating it, calculating reliability, and visualizing risks, while the final medical judgment and advice are always made by a pharmacist. This design aims to balance the convenience of AI technology with medical safety. [Explanation of Symbols]

[0068] 10...Client's device 11...Smartphone 12...PC device 13...Camera function 20...Pharmacist terminal 21...Administration screen 22...Consultation screen 23...Video call 100…Platform Server 110…Frontend 120…Backend 200... AI processing module group 210…AI Screening Module 220…AI Chat Assistant 230…Drug identification AI module 240…Interaction check module 250…AI Translation Module 260…AI Summary Module 270…Audit Logging Method 300...External AI API Integration Unit 400...Database 500...Payment processing unit 600...Video call processing unit 700...Drug database P100…Pharmacist List P200…Pharmacist Details P300…Appointment Booking P400...Consultation details P500...Medication identification AI P600...Medication record book P700...Interaction P800...AI Chat

Claims

1. A global pharmaceutical consultation support system comprising a consultation terminal, a pharmacist terminal, and a server device connected to the consultation terminal and the pharmacist terminal via a network, The server device is connected to a natural language processing AI model and an image recognition AI model via an external AI API integration unit. A symptom information receiving means that receives symptom information transmitted by the person seeking advice from the person seeking advice's terminal, An AI screening means that, using the natural language processing AI model connected via the external AI API linkage unit, analyzes the symptom information received by the symptom information receiving means to determine the optimal consultation category from a set of multiple consultation categories, extracts candidate pharmacists from the pharmacist database, and determines a recommended pharmacist based on a weighted overall score obtained from the pharmacist database, which includes an evaluation index, a consultation performance index, and a consultation type matching index for the candidate pharmacists. A drug image data receiving means that receives drug image data transmitted from the aforementioned client terminal, A drug identification means that, using the image recognition AI model connected via the external AI API linkage unit, extracts image features including color, shape, markings, and packaging information from the drug image data receiving means, retrieves drug information including the drug name, active ingredient, content, and manufacturer from a drug database that stores drug information from multiple countries, based on the extracted image features, generates information on equivalent drugs with matching active ingredients by referring to the drug database based on the retrieved drug information, and displays at least one of the following as the basis for selecting candidate equivalent drugs on the user terminal: matching active ingredients, matching or proximity of dosage, matching dosage form, and matching pharmacological classification. A means for receiving information on medications currently being taken by the client, which is transmitted from the client's terminal, An interaction check means that analyzes drug interactions based on the drug information and drug information received by the drug information receiving means, using the natural language processing AI model connected via the external AI API linkage unit, and determines the risk level based on at least one of the contraindications of the interaction, the need for dose adjustment, and the degree of caution required for concomitant use. A multilingual translation means that translates messages transmitted and received in consultation communication between the consultation terminal and the pharmacist terminal between multiple languages ​​by providing prompts to the natural language processing AI model, connected via the external AI API linkage unit, instructing it to perform translation processing corresponding to the specialized terms in order to ensure the accuracy of the translation of pharmaceutical terminology, cosmetic ingredient names, and health-related terms, thereby translating between multiple languages. A global pharmaceutical consultation and support system characterized by having the following features.

2. The global pharmaceutical consultation support system according to claim 1, further comprising an AI summary means that, after the consultation between the consultant and the pharmacist operating the pharmacist terminal is completed, analyzes the content of messages exchanged during the consultation using an AI model employing natural language processing, and extracts summary information and important matters.

3. The drug identification means receives image data of a drug acquired by the drug image data receiving means via the camera function or file upload function of the consultant terminal, extracts color, shape, markings and packaging information as image features from the image data and transmits them to an image recognition AI model, and generates information on equivalent drugs in at least two or more countries, including generic drug information, availability, estimated price range and precautions, based on the drug identification result returned from the image recognition AI model, as described in claim 1, for the global drug consultation support system.

4. The drug identification means calculates a confidence level based on at least one of the degree of matching of image features, the degree of matching of the marking recognition result, and the degree of matching of the packaging type for the processing result of drug identification by the image recognition AI model, and if the confidence level is less than a predetermined threshold, it sends an additional confirmation question to the user terminal, which includes at least one of inputting a marking string, instructing a reshoot, adding multiple angle images, and inputting packaging label characters, and performs reinference based on the answer to the additional confirmation question sent from the user terminal, as described in claim 1.

5. The global drug consultation support system according to claim 1, characterized in that the drug identification means, in generating equivalent drug candidate information, performs dose conversion based on different unit systems or standards for each country, and if there is a difference in dose or dosage form, adds warning information to that effect and the basis for the conversion and transmits it to the consultant terminal, and displays at least one of the following as the basis for selecting an equivalent drug candidate: matching active ingredient, matching or similar dose, matching dosage form, and matching therapeutic classification.

6. The interaction checking means, when additional information about a new drug is transmitted from the consultation terminal to the drug information being taken received by the drug information receiving means, retrieves a group of existing drugs with a "taking" flag set to true from the server device's database, performs differential evaluation of the interaction between the new drug and the group of existing drugs using an AI model employing natural language processing, determines the risk level of the interaction in five stages: severe, high, moderate, low, and none, transmits this to the consultation terminal along with a color-coded display according to the risk level, and generates a notification to guide the user to consult a pharmacist if the risk level is above a predetermined threshold, as described in claim 1.

7. The interaction checking means generates an uncertainty index in addition to a risk level as an evaluation result of the interaction, and highlights warning information when the uncertainty index is above a predetermined threshold, and the interaction evaluation uses matching results based on an external interaction database or rule engine in addition to estimation by a language model, and the recommended response information includes at least one of the following: recommendation to see a doctor, consultation to discontinue medication, consultation to adjust dosage, adjustment of administration time, and presentation of observation items, as described in claim 1.

8. The AI ​​screening means generates consultation type candidates and the confidence level of the candidates from the symptom information received by the symptom information receiving means, and if the confidence level is below a predetermined threshold, it sends an additional confirmation question to the consultation terminal including at least one of the following: whether or not the patient is taking medication, the purpose of the consultation, the urgency, the age group, and the medical history, to reclassify the patient, extracts candidate pharmacists from the database based on the determined consultation type, ranks the candidate pharmacists by a weighted overall score that includes an evaluation index, a consultation performance index, and a consultation type matching index, and sends the top N pharmacists and the basis for the ranking to the consultation terminal, characterized in that the global drug consultation support system according to claim 1.

9. The server device further includes an AI chat assistant means that responds 24 hours a day to questions about drugs and health using an AI model employing natural language processing via a chat widget floating on the display screen of the user's terminal, the AI ​​chat assistant means proposes the most appropriate consultation category and recommends an appropriate pharmacist based on the content of the user's question, generates a link to the pharmacist's profile page, and the chat widget is equipped with a quick reply button function, a conversation history retention function and a multi-language support function, as described in claim 1.

10. The global pharmaceutical consultation support system according to claim 1, characterized in that the multilingual translation means supports translation between at least five languages ​​including Japanese, English, Korean, Chinese, and Spanish, is equipped with prompt control to ensure translation accuracy specifically for pharmaceutical terminology, cosmetic ingredient names, and health-related terms, and accepts individual translation requests for each message on the consultation details screen of the consultation terminal and the pharmacist terminal.

11. The server device further includes a consultation management means that receives a consultation reservation request from the client terminal and manages a series of consultation flows including approval processing by a pharmacist operating the pharmacist terminal, online payment processing and video call processing, wherein the consultation management means sequentially executes a category selection step, a consultation content input step, a date and time selection step, a pharmacist approval step, a payment processing step, a video consultation implementation step and an AI summary generation step, and reflects the analysis results by the AI ​​screening means in the category selection step and the consultation content input step, characterized in that the global pharmaceutical consultation support system according to claim 1.

12. The server device further includes an audit log means for recording and storing various information transmitted from the consultant terminal, AI processing results, reliability, presented content transmitted to the consultant terminal or the pharmacist terminal, and the user's selection or additional response, along with the processing time, as an audit log for each of the means of the AI ​​screening means, the drug identification means, the interaction check means, and the multilingual translation means, and the audit log includes a record of the pharmacist's final confirmation of the processing results and is used to ensure traceability and quality control, characterized in that the global drug consultation support system according to claim 1.

13. A method for providing drug consultation support, which is performed by a consultation terminal, a pharmacist terminal, and a server device connected to the consultation terminal and the pharmacist terminal via a network, (a) An AI screening step in which symptom information transmitted from the consultation terminal is analyzed by a natural language processing AI model connected via an external AI API linkage unit, a candidate consultation type is determined from a plurality of pre-set consultation categories, a confidence score is generated which is a numerical value representing the validity of the determination result, if the confidence score is less than a predetermined threshold an additional confirmation question is sent to the consultation terminal and classification is performed again based on the answer to the additional confirmation question, candidate pharmacists are extracted from the pharmacist database based on the determined consultation type, and recommended pharmacists are ranked and sent to the consultation terminal based on weighted scores of a plurality of objective indicators including an evaluation index, a consultation performance index and a consultation type matching index obtained from the pharmacist database for the candidate pharmacists, and (b) Analyze the image data of the drug transmitted from the consultation terminal using an image recognition AI model connected via the external AI API linkage unit to extract image features including color, shape, markings and packaging information from the image data, and based on these image features, obtain drug information including the drug name, active ingredient, content and manufacturer from a drug database that stores drug information from multiple countries, and generate a confidence score based on the degree of matching of the image features, and if the confidence score is less than a predetermined threshold, send an additional confirmation question to the consultation terminal and perform inference again based on the answer to the additional confirmation question, and based on the confirmed drug information, refer to the drug database to generate equivalent drug information in multiple countries with matching active ingredients, along with country-specific regulatory classification information and dose conversion information, and display at least one of the following on the consultation terminal as the basis for selecting the candidate equivalent drug: matching active ingredient, matching or proximity of dose, matching dosage form and matching pharmacological classification, (c) When additional information on a new drug is transmitted from the user's terminal, the system retrieves existing drug groups whose "currently taking" flag is true from the server's database, and uses the natural language processing AI model connected via the external AI API linkage unit to perform a differential evaluation of the interaction between the new drug and the existing drug groups. The system generates a risk level and uncertainty index based on at least one of the contraindications of the interaction, the need for dose adjustment, and the degree of caution required for concomitant use. If the uncertainty index is above a predetermined threshold, the system highlights warning information and generates recommended response information including at least one of the following: recommendation to see a doctor, consultation on discontinuing medication, consultation on dose adjustment, adjustment of medication timing, and presentation of observation items. (d) A translation step in which messages in consultation communication between the consultation terminal and the pharmacist terminal are translated between multiple languages ​​by providing prompts to the natural language processing AI model connected via the external AI API linkage unit to instruct translation processing corresponding to the specialized terms in order to ensure the accuracy of the translation of specialized terms for pharmaceuticals, cosmetic ingredients, and health-related terms, (e) A logging step which records various information transmitted from the client terminal in each of the steps (a) to (d) above as an audit log, A method for supporting drug consultation, characterized in that the processing results by the AI ​​model in each of the steps (a) to (d) above are subject to final confirmation by the pharmacist operating the pharmacist terminal.

14. The drug consultation support method according to claim 13, further comprising the step of collecting the user's pharmacist selection results or satisfaction evaluations after consultation in the AI ​​screening step, and updating the weights or scoring rules of the objective indicators.

15. The global drug consultation support system according to claim 1, wherein the AI ​​screening means has an emergency sign detection unit that detects an emergency sign corresponding to a predetermined emergency symptom pattern from the symptom information received by the symptom information receiving means, and when the emergency sign detection unit detects an emergency sign, it transmits emergency warning information to the consultation terminal, which includes the content of the detected emergency symptom and emergency contact information corresponding to the country where the consultation terminal is located.

16. The interaction checking means includes a drug-food interaction analysis unit that analyzes interactions between drugs as well as interactions between drugs and foods or beverages, and the drug-food interaction analysis unit evaluates the interaction risk of the drug with predetermined food categories and transmits severity information, including the risk of emergency transport, and information on foods to be avoided to the consultant terminal, characterized in that it is a global drug consultation support system according to claim 1.

17. The global drug consultation support system according to claim 1, characterized in that the symptom information receiving means has a voice input function that converts voice data acquired via the microphone function of the consultation terminal into text by voice recognition processing and accepts it as symptom information.

18. A program for causing a computer to function as the AI ​​screening means, the drug identification means, the interaction checking means, the multilingual translation means, and the audit logging means in the global drug consultation support system described in claim 12. That's all.

Images